If CDs are mastered down to 16 bit, what good is a 24 bit DAC?

[arena]

My friend's band just finished recording an album, and during the final stages of mixing they mastered it down to 16 bit (I think with an interpolation or decimation filter, can't remember really) in order to put it onto compact discs, even though it was originally recorded in 24 bit.

But if all cds are mastered down to 16 bit, what good is a 24 bit DAC? Does it try to logarythmically predict or simulate the original 24 bit recording?

Also what exactly does 16 or 24 bit mean in terms of digital recordings?

Thanks

[MrMcGoo]

DVD-A and DTS disks can use the 24 bit "word" length. The Pioneer DV47Ai can pass a bitstream to the VSX49TXi to be decoded and do bass management for DVD-A and DTS as well as SACD. Some recordings seem to benefit from the newer formats IMHO. The Eagles' "Hell Freezes Over" is a good DTS example. Too bad there are not very many multi-channel recordings of such high quality yet.

Bill

[leok]

The # of bits determines the number of discrete voltage levels available between the most negative and positive waveform excursions. 16 bits define 65536 possible levels with 32768 being halfway between negative and positive, or what we usually think of as 0 Volts AC. 24 bits define 16777216 possible levels, which means the steps are much closer together and the final quantized waveform is smoother.

Using interpolation algorithms, digital "filters" try to reconstruct, from a CD's 16 bit information, steps that probably occured between the CD's 16 bit steps. The benefit, if it's done well, reduces noise and distortion that would occur directly decoding the 16 bit data.

There is an additional benefit of this interpolation, and that is, the sampling noise is shifted away from the audio range so that the analog filter required to remove it has very little impact on the final audio output.

leok

[Deang]

Is what you just explained known as "upsampling"? If so, thanks -- I've been wondering how that worked.

Tell us about your new player. I'm curious (as always) regarding your impressions of the lastest generation of DAC's.

[arena]

Don't know if you remember but I bought the Rega Planet 2000 to replace the Sony DVP-NS700P which I felt was harsh in the upper regions. My first impression was general surprise at how different the two players sounded. Initially I was disapapointed to realize I prefered the "in your face" sound of the Sony over what I thought was too much restraint or refinement in the Rega.

However, after the initial listen i "burned in" the Planet for 100 hours continuous according to a few online recommendations. I've Never really believed a component would play differently after 100 hours at operating temperature. But a few days after the Planet arrived I already heard a difference between the way the Rega sounded on the first listen and after about 50 hours of burn in.

So after burn in I sat down and put in Radiohead's 'Amnesiac' and was totally blown away. The whole sound had opened up, and the clarity and quality of tone and detail was unbelievable. Throughout the entire album I was consistently impressed by the detail and nuance the Planet pulled off the disc. I've never heard my speakers sound so good. The upper frequencies were especially impressive. The cymbals actually sounded like cymbals; they shimmered the way they are suppossed to, and each cymbal actually seemed to have an individual tone.

After 'Amnesiac' I put on Radiohead's 'OK Computer' and again heard details and nuance I've never heard before. This player has an amazing ability to draw you into the music.

I have no idea what happened, whether the player opened up after being burned in, or if I just warmed up to the Planet's sound, but I don't really care. For the first time I'm actually completely satisfied with my system. I don't even know if I'll upgrade my amp. I know I should, and I will hear improvements in doing so, but I don't feel the need to change anything right now, which is a really good feeling to have. Until know I would have summarized my feelings regarding my system in the following sentence:

"My system sounds fantastic. But"

But now I'm satisfied. At first I thought I would be selling the Rega, but it ended up selling me.

Anyway, the difference is very subtle, but definately there. It's not immediately noticible in direct a/b comparisons however, as the Sony seemed to reproduce all the same sounds the Rega did. But after repeat listening it became clear the overall tone and quality of the Rega was in a completely different class. Eventually I didn't even need to do the a/b comparisons, I could tell when listening to the Rega that I was hearing a quality of tone and detail I had never heard before.

Anyway, the new DACs should definately be checked out, just expect the improvements to be subtle and apparant only after repeat listenings.

[Griffinator]

Actually, the word length has nothing to do with the quality of the sound output. The samplerate, however, has everything to do with it.

A 192Khz D/A converter will typically produce just as many jitter errors as a 44.1Khz (CD rate) - but the 192Khz will do it well above the human hearing range (unless you are some superhuman with a system capable of reproducing frequencies around 96,000Hz) where the 44.1Khz D/A converter will do it right smack in the high frequency spectrum (around 18-22Khz) affecting the audible source.

That's the reason why even a cheap DVD player sounds better on CD playback than a garden variety consumer CD deck.

As far as bit rate is concerned, unless the deck is adding effects to the source signal, there is no benefit to 24 bit word length decoding on a 16 bit source.

[wireless]

----------------

On 5/28/2003 6:05:42 AM Griffinator wrote:

That's the reason why even a cheap DVD player sounds better on CD playback than a garden variety consumer CD deck.

----------------

Have you done many of these comparisons? I bought a Toshiba 2900 the other day with its prominently advertised 24 bit 192kHz chip. It did not sound good at all playing standard CDs. I had an old 80s Technics CD player and a computer with a SB live card and both were clearly better when doing A/B/C comparisons. It wasn't one of things were you sit back and try to pick out subtle differences. It just sounded bad.

There's a lot more circuitry other than the D/A chip. Even back when CD players were first being released, the sound quality all came back to the same thing - the quality of the components and filters. As any speaker designer or component design engineer will say, good ones are expensive and the sound is reflected therein.

[rowooo]

Griffinator...I'd like to know how you came to the conclusion that any cheap DVD player will better any garden variety CD player, this certainly isn't what I've found. My 1988 Delco am-fm car stereo sounds better than my Toshiba DVD player playing CD's. O.K. maybe it's not that bad but close.

[Griffinator]

----------------

On 5/28/2003 4:43:53 PM rowooo wrote:

Griffinator...I'd like to know how you came to the conclusion that any cheap DVD player will better any garden variety CD player, this certainly isn't what I've found. My 1988 Delco am-fm car stereo sounds better than my Toshiba DVD player playing CD's. O.K. maybe it's not that bad but close.

----------------

I did not say that - I said

"That's the reason why even a cheap DVD player sounds better on CD playback than a garden variety consumer CD deck."

Obviously there are exceptions to every rule.

I've done A/B comparisons with low-budget DVD players by GE, Sony, and even Mintek vs. standard consumer-grade CD players by Sony, JVC, etc. Every time, I found the high frequency response to be significantly clearer on the DVD players. This I can scientifically attribute to the D/A conversion issue.

Obviously, Toshiba has released a grade A dog if they couldn't do any better than a $50 player...

[leok]

Dean,

In a general sense, yes. Oversampling and upsampling, are algorithms that create samples in between the original 44.1KHz samples. In the end there are more samples per second. These algorithms can also create voltage values that are in between the 16 bit values that were originally placed on the disk. To utilize these in-between voltage values, one must use a DAC with in between values (usually a 20 or 24 bit DAC). Many implementations rearrange the whole thing by turning the sequence of original values into a series or plus or minus values that are cranked out at a very high rate (like 1MHz or so). There are lots of ways to reassemble the analog original (or close to it) from a 16 bit 44.1KHz digital representation.

leok

[ben.]

----------------

On 5/28/2003 6:05:42 AM Griffinator wrote:

Actually, the word length has nothing to do with the quality of the sound output. The samplerate, however, has everything to do with it.

----------------

Am I misunderstanding what you are trying to say, or are you on drugs? Ever hear an 8-bit sample? Ever dither a 24 bit clip down to 8 bit?

leok-

good nutshell explanation!

[Griffinator]

----------------

On 5/29/2003 1:44:55 PM bclarke421 wrote:

Am I misunderstanding what you are trying to say, or are you on drugs?

Ever hear an 8-bit sample? Ever dither a 24 bit clip down to 8 bit?

leok-

good nutshell explanation!

----------------

I think you're misunderstanding me (although I have been taking Lorcet for the last few days )

What I mean is that a 24 bit D/A converter will not do anything to improve the sound of a 16 bit signal.

[leok]

Dean,

I forgot to mention the Philips DVD963. I find that it is a big improvement over the Sonic Frontiers "TransDac" that I was using. The high end is much cleaner. Orchestra string sections often actually sound good. Sibilants are reproduced in much better condition. There is more detail and ambiance (better signal to noise ratio).

Note that this Philips device uses class A output amps. I think that's a big deal. Much better than the usual op-amps.

The SACD delivery is the only SACD I've heard. My fundamental impression is I can't hear any recording medium at all. It just sounds like whatever was recorded. So, I hear microphones and recording amps, and mostly,just the music.

leok

[ben.]

Griff-

That's what I was hoping! Thanks.

Ben

[leok]

There is one thing I thought of that hasn't been clear from my notes here, and may help to account for what might appear as a discrepancy between what I am saying and what Griffinator has been saying, when we have really been saying the same thing. Right.

Use of oversampling, upsampling, interpolation, 20-bit, 24-bit algorithms and hardware are tools in achieving theoretically perfect 16-bit d/a conversion. The end result will not be any better that the best theoretical 16-bit result. It's just that using a 16-bit DAC at 44.1K samples per second does not generally yield a satisfactory result. The algorithms and 20 to 24 bit hardware come closer to what is theoretically the best possible for 16 bits.

leok